The plot() Command

We can also examine the balance graphically using the plot() command, which provides three types of plots: jitter plots of the distance measure, Q-Q plots of each covariate, and histograms of the distance measure. For subclassification, separate Q-Q plots can be printed for each subclass. The jitter plot for subclassification is the same as that for other types of matching, with the addition of vertical lines indicating the subclass cut-points. With the histogram option, 4 histograms are provided: the original treated and control groups and the matched treated and control groups. For the Q-Q plots and the histograms, the weights that result after matching are used to create the plots.

Three examples of the output from the plot() command are shown in Figure . If the empirical distributions are the same in the treated and control groups, the points in the Q-Q plots would all lie on the 45 degree line (lower left panel of Figure ). Deviations from the 45 degree line indicate differences in the empirical distribution. The jitter plot (top panel) shows the overall distribution of propensity scores in the treated and control groups. In the jitter plot, which can be created by setting type = "jitter", the size of each point is proportional to the weight given to that unit. Observation names can be interactively identified by clicking the first mouse button near the units. The histograms (lower right panel) can be plotted by setting type = "hist".

Figure: Examples of the three types of output from the plot command resulting from matching on the /textttlalonde data set based on real earnings in 1974 (re74) divided by 1000, real earnings in 1975 (re75) divided by 1000, years of education (educ), Hispanic (hispan) and marital status (married). Observations in both the treated and the control groups outside the support of the distance measure were discarded. The upper plot shows the jitter plot of the distance measure. The lower left plot shows the QQ plots for the first three covariates (I(re74/1000), I(re75/1000), educ). The lower right plot shows the histograms of the density of propensity scores for observations before and after matching.